提取不存在的文件

Question

I have a function which I use to recursively rename files within a particular folder, like this:

public void renameRecur(string destination) 
{
    DirectoryInfo dirInfo = new DirectoryInfo(destination);

    //Rename each of the files
    foreach(FileInfo file in dirInfo.GetFiles()) 
    {
        file.IsReadOnly = false;

        /* Rename files and things... */
    }

    //Rename each of the directories
    foreach(DirectoryInfo dir in dirInfo.EnumerateDirectories()) 
    {
        renameRecur(dir.FullName);

        /* Rename directories and things... */
    }
}

However, file.IsReadOnly = false , will, on random occasion, throw a FileNotFoundException exception. After looking at the stack trace and information provided by the exception, I can use the provided full path and locate the "missing" file every time this exception occurs.

Again, this seems to occur on random occasions, usually whenever the application encounters a file that is in read-only mode, and it tries to make it writable. The program does not always break on the same file each time.

I am the administrator on the machine, and have full permissions to this folder and all of its sub-folders and files. None of these files or directories are hidden.

So, I'm curious as to why dirInfo.GetFiles() is returning files that C# thinks don't exist. Does anyone have any insight as to why this is happening?

Answer 1

Firstly, you don't have to implement recursion for this as the Directory.EnumerateFiles() takes a SearchOption parameter allowing you to include subdirectories. Not that this will solve your problem, but simpler code is always good, so thought I'd include the tip.

I think Adarsha (comment on your question) is onto something. But how can you use that to your advantage? I have two approaches in mind. There might be more..

1. Implement your logic so that it recovers gracefully when something bad like this occurrs. Most likely through one or more layers of retries and/or reruns. For example your inner code could retry once if the exception is caught. If it happens again, rethrow and let the consuming code handle it. Then let your consuming code rerun the whole loop after some sleep or whatever to let your file system settle down.

2. Use NTFS Transactions. Only available if your code will run on Vista/2008 or later. That stuff is in the windowsapi, so if you're uncomfortable with marshalling, you could use a prewritten wrapper such as Transactional NTFS over at codeplex. Not sure if you can lock your directories using that, but I think you can through Txf.Directory.GetDirectory(..) while providing your ambient transaction scope.

Best of luck! :)

Answer 2

I'm not really sure why DirectoryInfo will give you files that the system does not think exists; however, in order to help you with your exception. You should look into this :

http://msdn.microsoft.com/en-us/library/system.io.fileinfo.exists(v=vs.110).aspx

public void renameRecur(string destination) {
    DirectoryInfo dirInfo = new DirectoryInfo(destination);

    foreach(FileInfo file in dirInfo.getFiles()) {
        if(!file.Exists) //Try This.
            continue;
        file.IsReadOnly = false;

    }

    foreach(DirectoryInfo dir in dirInfo.EnumerateDirectories()) {
        renameRecur(dir.FullName);

    }
}

Edit based on comments:

You should take a look at this overload if you suspect that another part of the program has the file locked.

http://msdn.microsoft.com/en-us/library/5h0z48dh(v=vs.110).aspx

Here is an example from a past piece of code showing how you can use it (since I feel msdn lacks a meaningful example)

void ExternOpen(FileHeader header)
    {
        var appData = Environment.GetFolderPath(Environment.SpecialFolder.ApplicationData);
        var appDataLocation = appData + @"\" + header;
        using (var fs = new FileStream(appDataLocation, FileMode.Create, FileAccess.Write, FileShare.ReadWrite | FileShare.Delete))
        using (var hs = header.GetStream())
        {
            hs.CopyTo(fs);
            Process.Start(appDataLocation);
        }
    }

The code above gets a file that was created in memory via decompression (header.GetStream()) and then creates a physical file on the desk in the application Data and opens that file.

By specifying multiple FileShares I can allow for the file to be accessed in multiple ways.